Pacific Science (1991), vol. 45, no. 1: 28-49 © 1991 by University of Hawaii Press. All rights reserved Forest Structures, Composition, and Distribution on a Pacific Island, with Reference to Ecological Release and Speciation! YOSHIKAZU SHIMIZU2 AND HIDEO TABATA 3 ABSTRACT: Native forest and scrub of Chichijima, the largest island in the Bonins, were classified into five types based on structural features: Elaeocarpus­ Ardisia mesic forest, 13-16 m high, dominated by Elaeocarpus photiniaefolius and Ardisia sieboldii; Pinus-Schima mesic forest, 12-16 m high, consisting of Schima mertensiana and an introduced pine , Pinus lutchuensis; Rhaphiolepis­ Livistonia dry forest, 2-6 m high, mainly occupied by Rhaphiolepis indica v. integerrima; Distylium-Schima dry forest, 3-8 m high, dominated by Distylium lepidotum and Schima mertensiana; and Distylium-Pouteria dry scrub, 0.3­ 1.5 m high, mainly composed of Distylium lepidotum. A vegetation map based on this classification was developed. Species composition and structural features of each type were analyzed in terms of habitat condition and mechanisms of regeneration. A group of species such as Pouteria obovata, Syzgygium buxifo­ lium, Hibiscus glaber, Rhaphiolepis indica v. integerrima, and Pandanus boninen­ sis, all with different growth forms from large trees to stunted shrubs, was subdominant in all vegetation types. Schima mertensiana , an endemic pioneer tree, occurred in both secondary forests and climax forests as a dominant canopy species and may be an indication of "ecological release," a characteristic of oceanic islands with poor floras and little competitive pressure. Some taxonomic groups (Callicarpa, Symplocos, Pittosporum, etc.) have speciated in the under­ story of Distylium-Schima dry forest and Distylium-Pouteria dry scrub. Specia­ tion seems to have occurred exclusively where there are comparatively small numbers of component species, historically stable habitats, some opportunity for regeneration without large-scale disturbance, and the occasional occurrence ofcanopy gaps. THE BIOTA OF OCEANIC ISLANDS is derived from Mueller-Dombois 1981), Galapagos Islands ancestors that arrived by chance across an (Wiggins and Porter 1971, Eliasson 1984, open sea and became established. The fauna Hamann 1984), and Ponape, in the Caroline and flora of those islands have many char­ Islands (Nakamura 1985). Moreover, the acteristics that distinguish them from those of structure and quantitative composition of continents (see Carlquist 1974). There are plant communities have been analyzed with- many oceanic islands in the Pacific Ocean, but out consideration of insular phenomena such there is relatively little information about as ecological release and speciation. their vegetation except for the well-known The Bonin Islands (Ogasawara-shoto in Hawaiian Islands (Rock 1913, Fosberg 1961, Japanese), located about 1000 km south of Tokyo in the northwestern Pacific Ocean (Figure 1), are Japan's only typical oceanic I Manuscript accepted 10 April 1990. islands. They consist of about 30 islets occur­ 2 Department of Natural Sciences, The Komazawa ring in four main groups located linearly University, Koma zawa 1-23-1, Setagaya-ku, Tokyo 154, Japan. north to south: Mukojima, Chichijima, Haha­ 3 Laboratory for Plant Ecological Studies, Faculty of jima, and Iwojima. Even the largest island, Science, Kyoto University, Sakyo-ku , Kyoto 606, Japan. Chichijima, is comparatively small, 2395 ha in 28 Forest Structure on a Pacific Island-SHIMIZU AND TABATA 29 Chichi jima I. 20 -/--f--------,~----t----';--___j_ FIGURE 1. Location ofthe Bonin Islands and a geographical map ofChichijim a Island. Thin lines on the map show the IOO-m contour. Triangles mark principal mountains. area and 319 m highest elevation . The islands vegetation map for Chichijima. Okutomi et al. are volcanic, dating back to the Tertiary (1985) published vegetation maps for the (Mukojima, Chichijima, and Hahajima) or main islands of the Bonins. Ohba and Suga­ Quaternary (Iwojima) (Kuroda et al. 1981). wara (1977) and Okutomi et al. (1985) made The climate issubtropical. It rains almost year a detailed classification based on the Braun­ round, but soil water deficiency occurs in mid­ Blaunquet method of analysis. There are also summer in areas with shallow soil (Figure 2). ecological studies of the vegetation ofChichi­ Ninety-nine families, 256 genera, and 369 jima and Hahajima by Numata and Ohsawa species of vascular plants are found in the (1970) and Ohga et al. (1977). None of these Bonins. The proportion of endemism is 42.5% is sufficient, however, to describe a situation for all flowering plants and 74.7% for in which forest and scrub represent a stage on arboreal species alone (Kobayashi 1978). The which a drama of evolution is played. flora of the Bonins is related to that of sub­ In this paper we classify the native forest tropical Southeast Asia, Formosa, and the and scrub of Chichijima with emphasis on Ryukyu Islands (Hosokawa 1934, Tuyama structural characteristics and show some 1970, Yamazaki 1981). However, some gen­ unique features. We also attempt to relate era, such as Quercus and Castanopsis, which vegetation structure and composition to are the dominant species in the Ryukyu Is­ the phenomena of ecological release and lands, are absent in the Bonins (Shimizu, speciation. 1984a). Nomenclature follows Yamazaki (1970) Toyoda (1975) classified the vegetation except for Rhaphiolepis indica v. integerrima based on dominant species and proposed a and Cinnamomum insularimontanum. 30 PACIFIC SCIENCE, Volume 45, January 1991 1609mm (mm) 300 200 100 80 60 40 10 20 J FMAMJ J ASONDJ FIGURE 2. Climate diagram of Chichijima Island based on monthly mean temperatures and precipitation recorded from 1907to 1939. METHODS quadrats, randomly placed in each plot and covering at least 10% of the plot area . The native forest and scrub of Chichijima Profile diagrams were drafted, light inten­ was typified mainly by structural character­ sity was measured at ground level, and leaf istics derived from field observation. Aerial size (length and width) of all component photographs (scale 1 : 12,500)and ground ob­ species was measured with collected samples servations were combined to produce a vege­ in the typical forest and scrub ofeach type. tation map. Fifty plots, including each type of forest and scrub, were sampled in 1977 and 1979. RESULTS The forest vegetation was stratified into four height layers: Tl, the canopy tree layer; T2, Typification the low-stature tree layer; S, the shrub layer; and H , the herbaceous plant layer including The native forest and scrub on Chichijima seedlings of canopy species. Fewer layers were were classified into three main groups on the recognized in scrub. An emergent layer con­ basis of average height of canopy trees: mesic sisting exclusively of mature Pinus lutchuensis forest with tall trees, 10-15 m high; dry forest trees was designated the E layer. with low-stature trees, 2-10 m high; and dry All vascular plant species were enumerated. scrub with shrubs less than 2 m high. These Tree diameters (~3 em) were measured at groups were subdivided into fivetypes (Figure 1.2 m above ground except for shrubs less 3) in terms of structural features other than than 1.5 m high, for which stem girth was tree height (e.g., presence of an E layer and measured at ground level. Seedlings in the development of undergrowth shrubs). Types H layer and saplings ( < 3 em) were counted. were named by dominant or subdominant Herbs and vines were tallied in 1 m x 1 m species. A. E-A rn.f, C. R-L d.f. 1m O. 0-5 d.f. B. p-s m.f. E. O-P d.s. 1m 3mx 20m F IGURE 3. Profile diagrams of the five vegetatio n types. The size of sample strips is shown at bottom right of each. Keys to abbreviations: Ar, Ardisia sieboldii; Bl, Boninia glabra; Cy, Cyathea mertensiana; 01, Distylium lepidotum ; El, Elaeocarpus photiniaef olius; Hi, Hibiscus glaber; Ip , !lex percoriacea; Li, Ligustrum micrantum; Liv, Livistonia chinensis; Me, M elia azedarach; Pa, Pandanus boninensis; Pi, Pinus lutchuensis; Pis, Pisonia umbellifera; Po, Pouteria obovata ; Rh, Rhaphiolepis indica v. integerrima; Sb, Syzygium buxifolium; Sk, Symplocos kawakamii; Sm, Schima mertensiana ; Sp, Symplocos pergracilis; St, Sta chyurus praecox v. matsuzakii. 32 PACIFIC SCIENCE, Volume 45, Jan uary 1991 Elaeocarpus-Ardisia mesic forest (E-A m.f.) undergrowth shrubs occur in this kind of for­ is the tallest forest of all. Canopy trees attain est (Figure 3D). 15 m in height. Elaeocarpus photiniaefolius is Distylium-Pouteria dry scrub (D-P d.s.), of abundant in the canopy layer; lower layers are O.3-1.5-m-high vegetation, is also dominated dominated by Ardisia sieboldii (Figure 3A) . by Distylium lepidotum. Almost all species in Pinus-Schima mesic forest (P-S m.f.), 12­ this scrub have stunted growth forms (Figure 15 m high, is dominated by an introduced 3E). pine, Pinus lutchuensis, and an indigenous Figure 4 shows the similarity relationships pioneer tree, Schima mertensiana . The crowns of the sample plots based on the Bray-Curtis of the pine trees reach far above those of the (1957) ordination method. The index of simi­ other species, forming an emergent layer larity is IS = 200cj(a + b), in which a and b (Figure 3B) represent the total number of species in Plots Rhaphio lepis-Livistonia dry forest (R-L A and B, and c is the number of species com­ d.f.), 2-6 m high, is dominated by Rhaphio­ mon to both plots. The ordination, based on lepis indica v. integerrima (Figure 3C). There the composition of woody species, indicates are few shrub species in the S layer. that the plots of each vegetation type, except Distylium-Schima dry forest (D-S d.f.), 3­ those of the Pinus-Schima mesic forest, are 8 m high, is mainly composed of Distylium distinctively grouped. Pinus-Schima mesic lepidotum and Schima mertensiana. Many forest is a secondary forest (Shimizu 1983), 60 oI----r------r-----~..;::",.",,~ ~~--__r_--___r-----r--__,- 10 20 30 [,0 50 60 70 80 A x is 1 FIGURE 4.